Spatial gradients enable cells to precisely organize into complex multicellular patterns. Morphogens are released by clusters of cells and diffuse outward to form a gradient. Surrounding cells convert this gradient into patterns of gene expression using regulatory genetic programs composed of integrated genetic logic, feedback, and cell-cell communication. The central objective of this proposal is to identify the design principles by which genetic programs convert gradients into patterns of gene expression. Our approach will harness two new tools. The first is a set of red and green light sensors that activate a signaling pathway in a graded manner as a function of light intensity. These will be used to deliver light gradients that will be processed as inputs by a genetic program. The second is a suite of computational methods that enable us to convert a desired genetic program into a DNA sequence. This model-guided design allows us to enumerate and evaluate many genetic programs in silico and then construct and experimentally test a library of the most interesting programs. Further, a biophysical method that predicatively tunes expression levels by changing the ribosome binding site sequence will be harnessed to quantitatively compare the mutational robustness and evolvability of programs. The Aims are organized around the testing of hypotheses as to how regulatory networks process single and multiple gradient signals. Specifically, we will test the following theories: (i) Embedded feedforward loops improve robustness to mutations that affect expression levels (Aim 1). (ii) Feedback loops and spatial bistability increase pattern sharpness and robustness (Aim 1). (iii) Programs that integrate opposing gradients have improved robustness and pattern evolvability (Aim 2). (iv) Opposing gradients can be used as part of size-independent center-finding algorithms (Aim 2). The use of synthetic genetic circuits and E. coli as a model system will make it possible to characterize the gradient-processing properties of many genetic programs and make quantitative comparisons to a fully parameterized mathematical model. This approach will elucidate principles of how regulatory networks are organized to process gradients that can then be generalized to other systems.